Fix broken issue comes from OpenPAI API upgrade (#227)

Fix paiTrainingService broken issue comes from OpenPAI API upgrade
microsoft · Oct 17, 2018 · 42d8cbd · 42d8cbd
1 parent deaa187
commit 42d8cbd
Show file tree

Hide file tree

Showing 3 changed files with 19 additions and 12 deletions.
diff --git a/src/nni_manager/training_service/pai/paiData.ts b/src/nni_manager/training_service/pai/paiData.ts
@@ -69,4 +69,4 @@ export const PAI_OUTPUT_DIR_FORMAT: string =
 `hdfs://{0}:9000/`;
 
 export const PAI_LOG_PATH_FORMAT: string = 
-`http://{0}:50070/explorer.html#{1}`
+`http://{0}/webhdfs/explorer.html#{1}`
diff --git a/src/nni_manager/training_service/pai/paiJobInfoCollector.ts b/src/nni_manager/training_service/pai/paiJobInfoCollector.ts
@@ -69,7 +69,7 @@ export class PAIJobInfoCollector {
         // Rest call to get PAI job info and update status
         // Refer https://github.com/Microsoft/pai/blob/master/docs/rest-server/API.md for more detail about PAI Rest API
         const getJobInfoRequest: request.Options = {
-            uri: `http://${paiClusterConfig.host}:9186/api/v1/jobs/${paiTrialJob.paiJobName}`,
+            uri: `http://${paiClusterConfig.host}/rest-server/api/v1/user/${paiClusterConfig.userName}/jobs/${paiTrialJob.paiJobName}`,
             method: 'GET',
             json: true,
             headers: {

diff --git a/src/nni_manager/training_service/pai/paiTrainingService.ts b/src/nni_manager/training_service/pai/paiTrainingService.ts
@@ -244,7 +244,7 @@ class PAITrainingService implements TrainingService {
         // Step 3. Submit PAI job via Rest call
         // Refer https://github.com/Microsoft/pai/blob/master/docs/rest-server/API.md for more detail about PAI Rest API
         const submitJobRequest: request.Options = {
-            uri: `http://${this.paiClusterConfig.host}:9186/api/v1/jobs`,
+            uri: `http://${this.paiClusterConfig.host}/rest-server/api/v1/user/${this.paiClusterConfig.userName}/jobs`,
             method: 'POST',
             json: true,
             body: paiJobConfig,
@@ -291,7 +291,7 @@ class PAITrainingService implements TrainingService {
         }
 
         const stopJobRequest: request.Options = {
-            uri: `http://${this.paiClusterConfig.host}:9186/api/v1/jobs/${trialJobDetail.paiJobName}/executionType`,
+            uri: `http://${this.paiClusterConfig.host}/rest-server/api/v1/user/${this.paiClusterConfig.userName}/jobs/${trialJobDetail.paiJobName}/executionType`,
             method: 'PUT',
             json: true,
             body: {'value' : 'STOP'},
@@ -322,13 +322,15 @@ class PAITrainingService implements TrainingService {
 
                 this.hdfsClient = WebHDFS.createClient({
                     user: this.paiClusterConfig.userName,
-                    port: 50070,
+                    // Refer PAI document for Pylon mapping https://github.com/Microsoft/pai/tree/master/docs/pylon
+                    port: 80,
+                    path: '/webhdfs/webhdfs/v1',
                     host: this.paiClusterConfig.host
                 });
 
                 // Get PAI authentication token
                 const authentication_req: request.Options = {
-                    uri: `http://${this.paiClusterConfig.host}:9186/api/v1/token`,
+                    uri: `http://${this.paiClusterConfig.host}/rest-server/api/v1/token`,
                     method: 'POST',
                     json: true,
                     body: {
@@ -393,14 +395,19 @@ class PAITrainingService implements TrainingService {
                     this.hdfsBaseDir = "/";
                 }
 
-                const hdfsClient = WebHDFS.createClient({
-                    user: this.paiClusterConfig.userName,
-                    port: 50070,
-                    host: this.hdfsOutputHost
-                });
+                let dataOutputHdfsClient; 
+                if (this.paiClusterConfig.host === this.hdfsOutputHost && this.hdfsClient) {
+                    dataOutputHdfsClient = this.hdfsClient
+                } else {
+                    dataOutputHdfsClient = WebHDFS.createClient({
+                        user: this.paiClusterConfig.userName,
+                        port: 50070,
+                        host: this.hdfsOutputHost
+                    });
+                }
 
                 try {
-                    const exist : boolean = await HDFSClientUtility.pathExists("/", hdfsClient);
+                    const exist : boolean = await HDFSClientUtility.pathExists("/", dataOutputHdfsClient);
                     if(!exist) {
                         deferred.reject(new Error(`Please check hdfsOutputDir host!`));
                     }